class: center, middle, inverse, title-slide # Analytical Paleobiology ## with R ### Gregor H. Mathes ### University of Bayreuth/ Paleoecology Lectures ### 2021/06/17 (updated: 2021-06-16) --- # Overview .pull-left[ <br> **Today** - Tidy project structure - Troubleshooting - PBDB data - Binning of time series - Sample-standardization ] -- .pull-right[ <br> **Tomorrow** - Tidy data - import data - tibbles - Tidyverse - wrangle - iterate - visualize ] --- class:inverse, mline, center, middle # Tidy project structure --- # Project structure .center[] .footnote[ [Youtuber Rachelleea](https://www.youtube.com/channel/UCJCgaQzY5z5sA-xwkwibygw) ] --- # Project structure .pull-left[ <br> **Problem** - absolute paths - `rm(list = ls())` - non-reproducible results - total mess ] -- .pull-right[ <br> **Solution** - Rstudio projects - The `here` package - disenable workspace preservation ] --- # What is real? .center[<img src="data:image/png;base64,#https://d33wubrfki0l68.cloudfront.net/7fa44a5471d40025344176ede4169c5ad3159482/1577f/screenshots/rstudio-workspace.png" alt="Disenable workspace preservation in Rstudio" width="550"/> ] --- # R projects .center[] --- # R projects .center[] --- # R projects .center[] --- # R projects .center[] --- # Folder structures .center[] --- # Here Package <!-- .center[] --> .center[<img src="data:image/png;base64,#https://raw.githubusercontent.com/allisonhorst/stats-illustrations/master/rstats-artwork/here.png" alt="Allison Horsts visualisation of the here function" width="650"/> ] --- # Troubleshooting .center[] --- class:inverse, mline, center, middle # PBDB Data --- # Paleobiology Database [Download records](https://paleobiodb.org/classic/displayDownloadGenerator) - either directly via the website - or via the generated URL - both access the API to the database - carefully read the [documentation](https://paleobiodb.org/data1.2/) .pull-right[] --- # Binning observations .pull-left[ **Problem** - huge uncertainty around fossil ages - uneven spaces between occurrences ] .pull-right[ **Solution** - binning occurrences - assigning similarly aged fossils to a shared temporal unit - coarsen the resolution ] --- # Binning observations .center[ <!-- --> ] --- # Binning observations .center[ <!-- --> ] --- # Binning observations .panelset[ .panel[.panel-name[R Code] ```r library(divDyn) data("corals") corals %>% mutate(age_range = abs(max_ma - min_ma)) %>% ggplot(aes(x = age_range)) + geom_histogram(binwidth = 0.5) + labs(x = 'Age range (myr)', y = 'Count') + theme_minimal(base_size = 20) ``` ] .panel[.panel-name[Plot]  ] ] --- # Sample standardization .pull-left[ **Problem** - Data are samples - but we want to make inferences about populations - number of species increases with the number of individuals samples - comparison of collection of samples - invalid with different numbers of counted individuals ] .pull-right[ **Solution** - Sampling standardization - Previously: Classical rarefaction - Better: Shareholder Quorum Subsampling (SQS)¹ ] .footnote[ [1] [Alroy 2010](https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1475-4983.2010.01011.x) ] --- # SQS - an algorithm samples until taxa whose combined relative abundance exceeds a given threshold - threshold called the quorum - final results not dependent on quorum, but values between 0.3 and 0.7 are most commonly used - options to exclude singletons (species occuring only once in a sample) and dominants (the most abundant taxon) - only relevant for small data sets --- class:inverse, mline, center, middle # It's your turn --- # Task - set up a tidy project - go to the [PBDB](https://paleobiodb.org/classic/displayDownloadGenerator) and download some data you are interested in - either directly or just save the URL to load the data from within R - go through the `exercise_paleo.R` file <!--  --> .center[<img src="data:image/png;base64,#https://ak2.picdn.net/shutterstock/videos/15323182/thumb/8.jpg" alt="Allison Horsts visualisation of the here function" width="650"/> ]